Abstract

Exposed sites are highly vulnerable to decay, and shelters have been proposed as preventive conservation measures systematically. However, they need to be evaluated periodically as they may introduce new decay factors into the sites or provide an insufficient level of protection. Nonetheless, there is a lack of tools to help site managers make the complex decision of sheltering. Artificial intelligence systems can model information with a degree of uncertainty and are associated with material degradation processes and risky external hazards subjected to seasonality. This method has allowed a systematic assessment of the shelters at the Complutum archaeological site (Spain). The system has been applied twice a day during the summer, autumn, winter and spring seasons, taking into consideration atmospheric and en-vironmental input variables and vulnerabilities in the shelter design and management. By providing the model predictions, specific preventive conservation measures could be taken in advance to avoid further damage or failures.(c) 2022 Elsevier Masson SAS. All rights reserved.